Babylon Wrote the First Algorithm. We've Been Outsourcing Our Judgment Ever Since.
Somewhere in a climate-controlled museum case, there's a palm-sized clay tablet covered in cuneiform wedge marks. To most people it looks like a very old grocery list. To a computer scientist, it looks disturbingly familiar — because it is, functionally, a decision tree. If the grain stores are above X, allocate Y rations to Z workers. If not, reduce by this fraction, in this order, according to these priorities.
Babylonian bureaucrats were running rule-based systems roughly 4,000 years before anyone coined the word "algorithm." And the psychological instinct that drove them — the desire to remove human judgment from a complex, high-stakes decision and hand it to a reliable, consistent system — hasn't changed one bit.
The Original Logic Engine
The ancient Mesopotamian administrative state was, by any reasonable measure, a bureaucratic marvel. At its peak, the city of Ur was processing thousands of individual transactions per month across grain storage, labor allocation, livestock management, and temple economics. The scribes who managed this weren't just record-keepers. They were system designers.
Tablets recovered from sites across modern-day Iraq show standardized decision rules applied consistently across different administrators and different time periods. The rules didn't just record what happened — they determined what would happen. If a worker was absent for more than three days, this is what you deduct. If a field produced below this yield, this is how you reassign it. The human judgment had already been baked in, upstream, by whoever designed the system. The administrator on the ground was just executing the logic.
Sound familiar?
Your credit score works exactly the same way. Somebody, once, decided which variables matter and how much weight each one gets. Now that decision — made by a small group of humans at a specific moment in time, with specific assumptions — determines whether you get a mortgage, what interest rate you pay, and sometimes whether you get a job. You never meet the person who set the weights. You just live with the output.
The Comfort of the System
Here's the thing: algorithmic delegation isn't stupid. It solved a real problem in Babylon, and it solves real problems now. Human judgment is inconsistent, biased, and exhausting to apply at scale. A rule-based system is none of those things. It's predictable, it's fast, and — crucially — it diffuses blame. When the system says no, nobody said no. The system said no.
That last part is important. Historians studying Mesopotamian administrative records have noted that the rule-based approach served a political function as much as an operational one. When grain was scarce and rations had to be cut, a system that applied cuts according to pre-set rules was a lot easier to defend than a foreman who just decided who ate less this week. The algorithm was also a shield.
We do the same thing constantly. The loan officer doesn't deny your application — the algorithm does. The content moderation system removes your post — not any specific employee. The hiring software screens you out before a human ever sees your resume. The system becomes the responsible party, which means, in practice, nobody is.
When You Forget How to Decide
The Babylonian system worked well for a long time. And then it didn't, for a reason that should make anyone with a smartphone slightly uncomfortable.
When complex administrative systems become entrenched, the humans inside them gradually lose the capacity — and the institutional knowledge — to operate without them. This isn't speculation; it's a documented pattern across multiple collapsed administrative states. When the late Bronze Age palace economies fell apart around 1200 BCE, one of the cascading problems was that the scribal class who'd run the rule-based systems were gone, and nobody underneath them knew how to make the underlying decisions from scratch. The system had eaten the expertise.
Rome ran into a version of this with its late imperial tax collection bureaucracy. The rules became so elaborate, so layered with exceptions and sub-rules and workarounds, that the actual goal — collecting revenue to fund the state — got lost inside the machinery designed to achieve it. Administrators optimized for the system rather than the outcome. Sound familiar again?
In 2010, a flash crash on Wall Street wiped nearly a trillion dollars in market value in about 36 minutes, largely because automated trading systems were responding to each other faster than any human could intervene. Nobody decided to crash the market. The algorithms just did what they were designed to do, in conditions their designers hadn't anticipated. The humans in the room were mostly watching.
The Delegation Ratchet
There's a specific pattern that keeps showing up across history, and it's worth naming: call it the delegation ratchet. Systems get built to handle decisions that are too complex or too frequent for humans to manage manually. Those systems work, so they get trusted with more decisions. As they handle more decisions, human skill at making those decisions atrophies. Now the system is load-bearing, and you can't remove it even if you want to, because you've forgotten how to do what it does.
The Mesopotamian scribal system took about two centuries to become fully load-bearing. The US credit scoring system — FICO was introduced in 1989 — took about twenty years before banks genuinely couldn't function without it. The social media content algorithm took maybe ten years before the platforms themselves admitted they didn't fully understand what it was optimizing for.
The ratchet only turns one direction. Each click makes the system more essential and human judgment less practiced.
What the Record Actually Suggests
The historical record isn't an argument against algorithmic systems. It's an argument for maintaining what you might call decision literacy alongside them. The administrative states that survived disruption best — the ones that bent without breaking when the rules stopped working — were the ones that kept humans in the loop at enough points that the underlying judgment never fully atrophied.
The Venetian Republic ran a remarkably sophisticated administrative system for several centuries while deliberately building in human override mechanisms and rotating its decision-makers frequently enough that no single person, and no single rule set, became irreplaceable. It's not a coincidence that Venice lasted longer as a functioning state than almost any comparable power in history.
The question for anyone living inside an algorithmic system — which is everyone in America in 2024 — isn't whether to use the algorithm. It's whether you still know what you'd do without it. Because the Babylonian bureaucrat who built the first one probably thought the same thing: this is just a tool. We're still in charge.
They weren't, eventually. Neither, the record suggests, are we.
